Journal of Hydrology 389 (2010) 260–275

Contents lists available at ScienceDirect

Journal of Hydrology

journal homepage: www.elsevier .com/locate / jhydrol

Effects of Karst and geological structure on groundwater flow: The caseof Yarqon-Taninim Aquifer, Israel

Elad Dafny a,*,1, Avi Burg b, Haim Gvirtzman a

a Institute of Earth Sciences, The Hebrew University of Jerusalem, 91904 Jerusalem, Israelb Geological Survey of Israel, 95501 Jerusalem, Israel

a r t i c l e i n f o

Article history:Received 2 September 2009Received in revised form 6 May 2010Accepted 23 May 2010

This manuscript was handled by G. Syme,Editor-in-Chief, with the assistance of CraigT. Simmons, Associate Editor

Keywords:Judea Group AquiferFEFLOWYarqon-TaninimKarstMessinian Salinity CrisesPaleohydrology

0022-1694/$ - see front matter � 2010 Elsevier B.V. Adoi:10.1016/j.jhydrol.2010.05.038

* Corresponding author. Tel.: +972 2 6584272.E-mail address: Elad.dafny@mail.huji.ac.il (E. Dafn

1 Present address: Ecolog Engineering Ltd., Pekris 3,

s u m m a r y

This study demonstrates the significant influences of the geological structure (especially folding andlithology) and the karst system on groundwater flow regime. Folds divert groundwater flow from thegeneral hydraulic gradient; marly layers sustain several perched sub-aquifers above the regional aquifer;and karstification increases the hydraulic conductivity by several orders of magnitude. These phenomenaare quantitatively demonstrated within the Yarqon-Taninim (YT) basin, Israel, which is a complexgroundwater system, combining several (extremely) opposite characteristics: humid and arid rechargezones, phreatic and confined parts, shallow and deep sub-aquifers, stratified and relatively-homogeneoussub-basins, saline and fresh water bodies, as well as stagnant and fast-flowing groundwater regions.

We have introduced a 3D geological-based grid for the basin (for the first time). It was implementedinto a numerical code (FEFLOW), which was used thereafter to analyze quantitatively the flow regime,the groundwater mass balance, and the aquifer hydraulic properties. We present up to date conceptualunderstanding and numerical modeling of the YT flow field, especially at its mountainous parts.

Based on the calibration procedure and the sensitivity analyses, we obtained the best-fitted hydraulicconductivity values for the aquifer mesh. The general phenomenon observed is that as groundwater flowquantity increases, the hydraulic conductivity also increases. We interpret this result by the karstificationmechanism (including paleo-karst). Thus, where groundwater flow-lines converge and where groundwa-ter discharge amount increases, the karstification process intensifies and permeability increases. Conse-quently, at the mountainous region, along the syncline axes, where groundwater flow-lines converge,higher conductivities are found.

Modeling results also exhibit that at the lowland confined area, the geological structure does not play amajor role in directing groundwater flow. Rather, the flow field is controlled by the well-developed karstsystem and the relatively homogenous carbonate section. It is hypothesizes that the extensive karstifica-tion took place at the Messinian Salinity Crises, �5.5 Ma, during which groundwater heads as well as sealevel were lowered by several 100 m.

� 2010 Elsevier B.V. All rights reserved.

1. Introduction

1.1. The Yarqon-Taninim basin

The YT basin is one of Israel’s most important resources of freshwater, providing �25% of the countries fresh water (Hydrologicalservice, 2007). It stretches from the foothills of Mount Carmel inthe north to the northern Negev and Sinai anticlines in the south,and from the central mountain range in the east to the Mediterra-nean coastline in the west, over about 10,500 km2 (Fig. 1) (Golds-htoff and Shachnai, 1980; Shachnai, 1980). The basin is namedafter its two natural outlets: the Yarqon Springs, located at the cen-

ll rights reserved.

y).76702 Rehovot, Israel.

tral part of the basin, 15 km north-east of Tel-Aviv (altitude +18 masl), and the Taninim Springs, located at the northern tip of the ba-sin (+3 m asl).

The YT comprises of a western confined part and an eastern,mountainous phreatic part, where replenishment occurs. Climatein the replenishment area is of a Mediterranean one, with an aver-age precipitation of 550–600 mm/year, during the winter. The esti-mated recharge ranges between 330 and 360 million cubic metersper year (MCM/year), which accounts for 30–33% of the annualprecipitation on the average (Guttman and Zukerman, 1995). Inaddition, 3–5 MCM/year of deep-seated saline water, having asalinity of seawater, enters to the bottom of the aquifer, at thenorthern edge of the basin, producing the brackish groundwaterat the northern Taninim outlets (Paster et al., 2005). Groundwaterheads at the mountainous area reach 660 m in Hebron, 500 m inJerusalem and 300 m in Samaria. At the lowland it gently drops

BA

B

35o

30o

35o

Fig. 1. Location maps. The indicated cross-sections are shown in Fig. 3. Abbreviations (from south to north): MS – Mash’abey-Sade, AV – Arad Valley, MA – Ma’on anticline,HA – Hebron anticline, EK – En Karem basin, KM – Kefar Menahem well, RA – Ram’alla anticline, KS – Kefar Saba, AA – Anabta anticline, T/1 – Menashe T/1 well, YA – Yironanticline, MH – Menashe Height.

E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275 261

from south to north: at the 1930s, referred here as the ‘historicalperiod’, before pumping had started, heads dropped from 27 mnear Beer-Sheva at the south to 16 m in the Meanashe Heights atthe north.

At the historical period, the YT groundwater drained toward thetwo major springs and additional minor outlets near the TaninimSprings (Table 1). Since the 1950s, the basin is exploited inten-sively by hundred of wells for domestic and agriculture purposes.During the last decade, the YT supplied about 405 MCM/year onthe average, of which 20–30 MCM/year through the Taninim

Springs (�25% of its original discharge) and the rest through wells(Hydrological service, 2007). Due to the overexploitation waterlevels at the confined part of the aquifer dropped by 15–10 mand the Yarqon Springs totally dried since 1962.

1.2. Geological settings

The YT groundwater flows within the Judea Group Aquifer(JGA). The Late Cretaceous Judea Group is mostly composed of kar-stic, permeable limestone and dolomite interbedded with thin ter-

Table 1The YT water budget (MCM/year).

Historical, pre utilizationconditions

Current conditions

Assumed/measured Calculated

OutputsYarqon Springs 226–228 227.6 DryTaninim OutletsTaninim Springs 91–93 93 20–40Shuni Springs 6–7 6.5 DryLeak to Caesareaa 2–3 2.5 2–3Mediterranean 2.5–3.5 2.4 2.5–3.5

Total 328–335 332 24–47+ 308–405 pumps

InputsRecharge 323–332 327.5Lower saline flushing 3–5 4.5Total 328–335 332

a Guttman (2002).

262 E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275

rigenous clays and marls (Arkin and Braun, 1965; Arkin and Hama-oui, 1967). It was deposited on a wide and shallow carbonatic plat-form, governs a continental environment at the east and south(Lewy, 1991). Thus the lithology of the Judea Group tends to behomogenous at the west, with an increasing proportion of terrige-nous beds landward (Fig. 2A). A gradual lateral transition from thepermeable Judea Group rocks (the platform facies) to the mainlyargillaceous and non-permeable Talme Yafe Group (the slope fa-cies), which is its synchronous equivalent, occurs at the westernrim of the country, approximately along the Mediterranean shore-line (Bein, 1974).

The Judea Group overlies the Kurnub Group. For the same rea-son, facies at the upper part of the Kurnub Group changes fromargillaceous limestone of the Yakhini Formation at the west tomarls of the Qatana and Tamun formations at the east (Figs. 2A,3D–F) (Fleischer et al., 1993). The Judea Group is overlain by chalkyrocks of Mount Scopus Group, or in places where the last is miss-ing, by marly Neogene units. The thickness of the Judea Group in-creases gradually from about 500 m at the northern Negev to about800–1000 m in the central and northern parts of the basin.

A detailed division of the Judea Groups into ten formations wassuggested for the mountain region (Arkin and Hamaoui, 1967)(Fig. 2A), and later was extended to other portions of the basin(Ben Gai et al., 2007). Six formations, namely Kefira, Gi’vat Ye’arim,Kesalon, Amminadav, Weradim and Bina, are comprise of purelimestone which had partly or fully re-crystallized to dolomitesoon after deposition (Sass and Katz, 1982). Soreq and Bet Me’ir

Table 2Summary of previous numerical models for the YT basin.

Reference Modeleddomain

Type(FD/FE)

No.layers

Phreatic Confi

Baida et al. (1978) South + Hebron FD 1 v vGoldshtoff and Shachnai (1980) Entire FD 1 vMeiri and Guttman (1984) Jerusalem Mts. FE 1 vGuttman (1991) South FE 1 vBaida and Zukerman (1992) Jerusalem Mts. FD 1 vGuttman and Zukerman (1995) Entire FD 1 v

North FE 1 va v

Guttman and Zeitoun (1996) North FE 2 vBerger (1999) Entire FE 1 v vThis study Entire FE 3 v v

Abbriviations: FD – Finite Differences (e.g., MODFLOW); FE – Finite Elements (e.g., FETransmissivity; n – porosity; a – diffusive coefficient.

a SUTRA deals with phreatic conditions; modeling of the confined part was conducted

formations are comprise of dolomites interbedded with marls.Moza Formation comprises a marly layer. All these formations ex-posed over almost the entire mountain area and lie in conformityone above the other. The chalky Kefar Sha’ul Formation was depos-ited within local troughs and basins (Sass and Katz, 1982), thus itsthickness is considerably vary between 0 and 150 m and its faciesshifts in places to chalky limestone.

The current structure of the Judea Group rocks was shapedduring two stages: the Senonian-Eocene Syrian-arc folding event,at which several en-echelon NE–NNE trending anticlines wereformed (Garfunkel, 1988), and younger uplift events, at whichthe Judea Group was exposed and eroded (Bar et al., 2006).The latter derived sediments were transported westward, anddeposited at the lowland. Currently, most of the anticlines areburied beneath these younger units, while some exposed andcomprise the central mountains range of Israel. The notable ofwhich, are the Anabta, Ram’alla, Hebron and Ma’on anticlines(Figs. 1 and 3).

In some places along the western paleo-continent slopes, deepcanyons carved the Judea Group rocks since the Oligocene and dur-ing the Neogene (Druckman et al., 1995). Later, at the MessinianSalinity Crisis, when the Mediterranean Sea desiccated, groundwa-ter drained through these canyons (Gvirtzman, 2002). Neogenemarls accumulated above the previous topography, filled the can-yons and reached a thickness of about 400 m (and much thickerbeneath the current sea floor).

1.3. Hydrogeology

At the confined part of the basin, where most wells exist, karstis well-developed and groundwater flow is driven by head gradi-ent. The karst nature of the JGA is indicated by plenty of directand indirect observations: Enormous transmissivities, up to105 m2/day were reported from pumping test, as well as negligibledrawdowns (less than 1 cm) during pumpings of up to 103 m3/h;Caves and cavernous were discovered during TV profiling of severalwells where others are inferred from drilling-bit drops reported inthe drillings log and geophysical logs (Ben Gai et al., 2007); Mazekarstic caves of different sizes were discovered and studied inten-sively in the upper, exposed part of the aquifer (Frumkin and Fisch-hendler, 2005; Frumkin and Gvirtzman, 2006).

Traditionally, the JGA is divided into upper and lower sub-aqui-fers by the intermediate leaky aquitard of Moza and Bet-Me’ir for-mations (Fig. 2b) (Shachnai, 1980; Mercado, 1980). The leakagewas inferred from water balance considerations: large volume ofrecharge feeds the lower sub-aquifer over the mountainous area,while all the natural outlets and most pumping wells, located atthe lowland, are fed by the upper one.

ned Time steps(months)

Calibratedperiod

Calibratedparameters

Remarks

6 1956–1976 S, T6 1952–1976 S, T

1982, Static T6 1976–1990 S, T6 1967–1989 S, T As part of the Mts. aquifer6 1952–1993 S, T

n, at, al 2D SUTRA, flow and transport

6 1971–1994 T, at, al Flat geometry, flow and transport1 1988–1994 S, T +Hydrometeorological model1 1987–2003 K, S0 +Hydrometeorological model

FLOW); K – Hydraulic conductivity; S – Storativity; S0 – Specific storativity; T –

by several manipulations.

HebronJerusalem& Binyamin

Samaria LowlandConvention(Geology)

low

erup

per

mid

B

A S

Turo

nian

Bi'na

Weradim

Soreq

Kefira

Bet Me'ir

'Amminadav

Kesalon

Qatana

Giv'at Ye'arim

Moza

Kefar Sha'ul

Fig. 2. Stratigraphic (A) and hydrogeological (B) schemes for the Judea Group: (A) note the seaward-landward facies changes and (B) the colors of the conventional divisioncolumn represent the geological layers, as shown also in all cross-sections (Fig. 3). Lithostratigraphy after Arkin, 1976. The other columns show local sub-division of the JGA:the saturate zone (in light blue), the water table and the aquitards (black). (For interpretation of the references to colour in this figure legend, the reader is referred to the webversion of this article.)

E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275 263

The flow within the confined part is conventionally presentedon a north–south head profile (Fig. 4). Since the Yarqon Springsdried out, the general flow gradient is northward. At the historicalperiod however, the YT basin was divided by internal water divideinto two sub-basins, corresponding to its two natural outlets. Thewater divide was recognized up to the early 1960s, few kilometersnorth of the Yarqon Springs, as a local head maximum (Fig. 4) (e.g.,Mandel, 1961). The water divide separates the flow toward eitherof the two outlets. The observed hydraulic gradient is exceedingnorthward from 0.02‰ to 0.03‰ between Beer Sheva and the Yar-

qon Springs, to about 0.10‰ between the Yarqon Springs to Ma’a-nit and to about 1‰ between Ma’anit and the Taninim Springs(Mandel, 1961; Weinberger et al., 1994; Hydrological service,2007).

The east–west head gradient at the confined part, if existed atall, is obviously much less, and cannot be precisely detected be-cause most of the wells are located along the foothills (a narrowbelt along the Samaria and Judea Mountains). At the few excep-tions, measured heads at the lowland, at the west and at the foot-hills at the east, are resemble.

Fig. 3. Geological cross-sections: (A) along the confined part of the YT basin, vertical exaggeration 10�; (B) Arad Syncline, vertical exaggeration 3�; (C) Hebron Mountains;(D) Jerusalem Mountains; (E) Binyamin Mountains; and (F) Samaria Mountains. Locations of cross-sections are marked at Fig. 1. Vertical exaggeration 5�, except otherwisestated. Abbreviation (from south to north): MS – Mash’abey-Sade, BS – Beer Sheva, KG – Kiriyat Gat, H – Hadera, T – Tireli Fault.

264 E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275

At the phreatic parts, the groundwater flow field is diverted bythe major folds (Fig. 5). The karst is less developed: Transmissivi-

ties reach only about 103 m2/day (Baida et al., 1978; Baida andZukerman, 1992) and the observed caves are of discrete ‘chamber

Fig. 3 (continued)

E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275 265

type’ (Frumkin and Fischhendler, 2005) and hydrostratigraphydivision of the JGA varies regionally, depending on the local stratig-raphy (from south to north):

Four sub-aquifers appear within the JGA in the Hebron Moun-tains (Fig. 2B) (Kolton, 1973; Guttman and Gotleib, 1996). Thetwo upper ones spread un-continuously over restricted area andfeed different perched springs and four wells. However, most ofthe water leaks downwards to the two regional lower ones, andflow down-gradient in-between the Hebron and Ma’on anticlines.Average head gradient is about 13‰ (325 m/25 km).

At the Jerusalem Mountains area, only the lower part of the JGAis saturated and exploited by wells (Fig. 3D). The upper sub-aqui-fers are not continuous at all; rather many small pieces of perchedaquifers feed small springs at two major horizons (Fig. 2B) (Weissand Gvirtzman, 2007). Most groundwater at the lower regionalaquifer flow to the south-west, in-between Ram’alla and Hebronanticlines (Meiri and Guttman, 1984; Baida and Zukerman,1992). Structural boreholes drilled upon the western flank of theRam’alla anticline reveal a thin saturated section at the base ofJGA, indicating additional flow-paths of groundwater across the

MS BS G A

KS

Yarqon

Taninim

30

25

20

15

10

5

0

Hea

d [m

]Simulated

Measured

Spring

Distance from Mash'abey-Sade [Km] 20 40 60 80 100 120 140 160 1800

Fig. 4. N–S head profile along the confined part of the YT basin for the historical, pre-production period. Dots – measured/estimated levels at wells; line – simulated. Profileslocation is along cross-section A (Fig. 3).

266 E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275

anticline. Average head gradients are of about 22‰ (475 m/21 km)parallel to- and 82‰ (450 m/5.5 km), oblique to the anticline axis.

Similar division is inferred at the Binyamin Mountains, based onthe occurrence of many small perched springs. It appears that onlythe lower part of the JGA is saturated, and that heads are increasingeastward, but in a moderate slope compare to the aquifer base;thus the saturate section thins to zero and the Ram’alla anticlineact as a physical barrier to groundwater flow (Fig. 3E).

At most of the Samaria Mountains, only the lower part of theJGA is saturated. At restricted areas, a perched aquifer exists abovethe Moza and Bet Me’ir formations. Groundwater flows from thehighest area near Shilo to the west and northwest (Fig. 5). At his-torical times, when the Yarqon Springs emerged, an internalwater-divide cut through the Samaria, separated the Yarqon sub-basin from the Taninim sub-basin.

1.4. Objectives

Our objective is to study the influences of the geological struc-ture (especially folding and lithology) and the karst system ongroundwater flow regime. We choose the case of the YT basin, Is-rael, since it is well-documented and studied (e.g., Weinbergeret al., 1994) and since detailed structural maps were lately accom-plish (Ben Gai et al., 2007).

2. Numerical model

The numerical model is introduced using the FEFLOW code(Version 5.2) of WASY’s GmbH (Diersch, 1996). FEFLOW is a com-mercial finite-element model which is able to address the uniquespecification needed for the YT basin including a complex topogra-phy and structure, multiply water tables and lateral shift from con-fined to unconfined conditions. The FEFLOW solves the continuumequations for liquids (i.e. water) in porous media (Diersch, 1996).We assumed it is appropriate to model the JGA karstic aquiferusing the ‘Equivalent continuum approach’, because both thegroundwater flow within the matrix and through karst conduitscan be averaged into a bulk conductivity of the model’s cells. Thisapproach was found to be appropriate for well-connected fracturesystems at a fairly large scale (Ford and Williams, 2007).

Further FEFLOW definitions within the text are italic. Completetechnical definitions are listed in Appendix 1 in Supplementarymaterial.

2.1. Discretization

The models mesh comprises of three layers, each having 7395triangle prisms elements. Elements dimensions varied according

to geological considerations and locations of nearby wells(Fig. 6). The layers represent the three main hydrogeological units,i.e. the upper sub-aquifer, the intermediate aquitard and the lowersub-aquifer (Fig. 2B at left). The general structure of the model,including the layers configuration is demonstrated at Figs. 3 and7. At places where the upper units eroded or unsaturated, all threelayers represent the lower sub-aquifer. Slice elevations were digi-tized, using ESRI ArcGIS software, using new structural maps ofthe relevant horizons within Judea Group (Ben Gai et al., 2007).

Two exceptions of the above division exist. At the Taninim blockthe upper layer represents the sandy Pleistocene Kurkar Group di-rectly overlies the JGA, and the 3rd layer represent the Talme YafeAquiclude; At the Hebron Mountains the intermediate aquitardwas set to describe the Soreq Formation, separating the Kesalonsub-aquifer above it and Kefira sub-aquifer below it (Kolton, 1973).

2.2. Boundary conditions

The outer boundary of the model was delineated along geolog-ical structures or through well-defined water divides which can bedescribed as ‘no flow’ boundaries. Segments which serve as outletswere set as other boundaries as follow (Fig. 6):

A 1st kind boundary condition (Dirichlet type) was appliedalong the 1.5 km shoreline of the Taninim elevated block (‘C’ atFig. 6B), the only place where the YT basin is connected to the Med-iterranean Sea. The boundary condition prescribes a hydrostaticcolumn by accounting the increased pressure under the saline,dense seawater:

hðzÞ ¼ 0þ z � Cf � Cs

Cf

� �ð1Þ

where h(z) is the hydraulic freshwater head at elevation z above sealevel and Cf and Cs are the densities of the fresh and saline water,respectively.

A 3rd kind boundary condition (Cauchy type) was set at thesprings which drain the basin, i.e. the Yarqon, Taninim and Shunisprings:

q ¼ �UhðhR � hÞ ð2Þ

The boundary threshold head (hR) was set to the topographiclevel near the springs, i.e. 18 m, 2.5 m and 9 m, respectively, whilethe ‘transfer coefficient’ (Uh) was set manually during the calibra-tion process to meet the observed flux (q). The spring’s cells wereconstrained to allow only outflow, while disallow inflow at timeswhen the surrounding heads drop below the threshold head.

A 4th kind boundary condition (‘single well’) is applied along thesouthern part of the Taninim elevated block (dotted line at Fig. 6B),through which a small leak occur southward. The cumulative

B

C

A

Head (1m intervals at the lowland) Head (50m intervals at the mountains)

175000 200000 200000150000 175000

5650

00

5650

00

5900

00

5900

00

6150

00

6150

0061

5000

6150

00

6400

00

6400

00

6400

00

6400

00

6650

00

6650

00

6650

00

6650

00

6900

00

6900

00

7150

00

7150

0071

5000

7150

00200000

200000

200000

200000

175000

175000

150000

Legend

Fig. 5. Calculated groundwater flow field under historical pre-production conditions: (A) the entire basin; (B) enlargement of the northern edge; and (C) enlargement of thecentral part. Note the different intervals of the head contours. Encircled numbers are spring discharges in MCM/year.

E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275 267

discharge through this segment was 2.5 MCM/year, to fit previousestimations (Guttman, 2002).

Recharge (‘inflow at top’) was set as a fraction of the average an-nual rain over the replenishments areas. The recharge rate wasfound to be 33–34% at the Mediterranean climate, 25% at thesemi-arid climate and 5% at the arid climate. These rates satisfy

water budgets of the entire basin and the two sub-basins (Table 1).The 3–5 MCM/year saline water source entering the basin near theTaninim outlet, was represented by ‘Inflow from the bottom’ (‘E’ atFig. 6B), according to conceptual understanding: the saline ground-water leached from the bottom next to the fault zone, but not nec-essarily through the faults. This idea is represented graphically at

Fig. 6. A full 3D mesh of the YT basin (A) and an enlargement of the grid at the northern edge (B). The top surface colors are shaded relief. Abbreviations: AA – Anabtaanticline; B – Binyamina Fault; C – Mediterranean coast; E – Encroachment of saline water from bottom; HA – Hebron anticline; MA – Ma’on anticline; RA – Ram’allaanticline; S – Shuni Spring; T – Tireli Fault.

268 E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275

Fig. 3A (right side), where the northward flow-lines converge andredirect upwards to the Taninim Springs. At this point the salinegroundwater is leached from the bottom.

2.3. Calibration

Calibration goal was to reconstruct the natural, historicalgroundwater flow field which prevails at the YT basin beforegroundwater exploitation started. For this purpose a steady-statemodel was applied with the averaged water balance, presentedabove (Table 1). However, unsteady numerical solutions occurredas cells along the anticlines, which essentially should be saturated,were dry out. These resulted from the steep topography of the lay-ers (Rühaak W., WASY GmbH., Personal Communication, 2007).This was overcome by setting dynamic conditions (which includea storage term) and enable the heads to gradually adjust to anychange in the conductivity or other calibrated parameters, untilsteady-state is achieved. On the other hand, setting dynamic con-dition made the automatic calibration (PEST) time-consuming,un-robust and inconsistent thus unpractical.

Calibration was conducted by varying the ‘transfer coefficient’ atthe springs and the horizontal hydraulic conductivity of the layers.

Initially, one conductivity value was assigned to the confined zoneand one to the phreatic zone. Once a reasonable solution reached,the layers were divided into areas of varying conductivity for finetuning. To reduce the numerous degrees of freedom, three essen-tial suppositions were followed:

(1) A 1:10 vertical anisotropy (Kx = Ky = 10Kz). Previously studiesprescribed the anisotropy of the JGA 1:5-1:100 (Matmon,1995; Rimmer, 2003; Laronne Ben-Itzhak and Gvirtzman,2005), the lower ratios assumed to result by the high con-nectivity of the karst conduits.

(2) A 1:1 ratio between the upper and lower sub-aquifers. Thissupposition is applicable in a lack of contradicting field-data,and in order to avoid imprudent effects which can be carriedotherwise, such as diverting the flow to the relatively high-conductance sub-aquifer (e.g., Freeze and Witherspoon,1967).

(3) A 1:10 ratio between the aquifer units and the allegedlyintermediate aquitard at the confined zone. This was setmerely to represent the increased marls content within theaquitard. However, in places where the intermediate layerrepresent an efficient aquitard, its hydraulic conductivity,

NW

En Karem 8“Rail wells”

Modi’in 1

En Karem 13

Qatana Fm.

enilcitnaRam’alla

SE

Zif 1

Hebron

enilcitna

SE

Qatana Fm.

enMa’on

ilcitna

Yatta 2

NW

A

B

C

+30

Shibteen 5

+23

Alishama

anticlineRam’alla

Fig. 7. Hydrogeological cross-sections of: (A) Hebron Mountains; (B) Jerusalem Mountains; and (C) Binyamin Mountains. These cross-sections show the simulated saturatedzones (light blue) and are corresponding to the geological cross-sections at Fig. 3C–E, respectively. Vertical exaggeration: 5�. (For interpretation of the references to colour inthis figure legend, the reader is referred to the web version of this article.)

E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275 269

as wells as the vertical anisotropy, were manually altered toachieve fine calibration.

Head calibration was based upon 56 wells spreading through-out the entire basin (Table 3, Fig. 5). For each, a representative headwas carefully set. The wells were clustered into six groups, accord-ing to their geographical location and a zonal Root Mean Square Er-ror (zonal RMSE) was calculated:

Zonal RMSE ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1n

Xn

i¼1ðOi � CiÞ2

rð3Þ

where n is the number of the wells in each group, Oi is the observedhead and Ci is the calculated head.

Calibration concluded when the discrepancies between calcu-lated and the measured/assumed heads at each zone and dis-charges at each outlet were minimal, in accordance withmeasurements reliability.

The natural, historical head map was satisfactorily recon-structed (Fig. 5). The N–S head profile across the confined zone ispresented also in Fig. 4. General agreement exists between thesimulated and the measured heads at both phreatic and confinedzones (Fig. 8). The discrepancy at single wells varies between�5 m and +5 m with an average of 1.1 m and the zonal RMSE variesbetween 0.8 m and 2.7 m (Table 3).

2.4. Hydraulic conductivity variations

The calibrated horizontal hydraulic conductivity of the threehydrogeological layers is shown in Fig. 9. Conductivity of the alleg-edly sub-aquifers (A and C at Fig. 9) was found to vary between10�4 m/day and 103 m/day (Tables 4–7). These values are typicalof karstic limestone (Freeze and Cherry, 1979). Low conductivitiesof 10�2–1 m/day characterize the mountainous area and areabruptly changed to high conductivities of 102 m/day at the low-land. Exceptional conductivity of 103 m/day characterizes the Tan-

Table 3Measured vs. calculated heads for the historical, pre-developed, conditions (m).

Zone Well Observed/assumed head Uncertainty Calculated head Difference Zonal RMSE

Hebron Mountains Fawar 3 657.0 10 654.56 �2.44

3.4Yata 2 533.0 2 531.86 �1.14Samoa 522.0 2 522.86 0.86Zif 1(Kesalon) 570.0 5 563.28 �6.72Zif 1 350.0 20 352.12 2.12

Jerusalem Mountains Bar Giora 214.0 1 214.57 0.57

2.4

EK 11 411.0 2 407.94 �3.06EK 8 487.0 2 482.62 �4.38EK 2 485.0 2 489.08 4.08EK 13 445.0 2 446.17 1.17EK 12 502.0 0.5 501.64 �0.36EK 7 397.0 0.5 396.73 �0.27EK 5 464.0 0.5 463.35 �0.65

Negev Mashabe Sade 2 26.5 0.1 26.46 �0.04

0.1

Arad 4A 26.7 0.1 26.62 �0.08Ze’elim 1 26.4 0.1 26.36 �0.04Beer Sheva obs. 26.4 0.1 26.38 �0.02Tel Shoqet 1 26.7 0.1 26.59 �0.11Nevatim 1 26.7 0.1 26.57 �0.13Nizzana 1 26.4 0.5 26.40 0.00Arad 2 123.0 0.2 123.26 0.26

Northern edge Tireli 6 14.3 1 15.30 1.00

0.5

Taninim T/5 4.2 0.1 4.09 �0.11Shuni 2 7.8 0.5 7.72 �0.08Binyamina 5 9.6 0.5 9.41 �0.19Binyamina 1 6.10 0.5 6.22 0.12Taninim T/2 5.50 0.2 5.57 0.07U Binyamina 5.1 0.2 5.87 0.77

Samaria and Binyamin Haris 280.0 3 283.83 3.83

1.7

Shibtin 1 93.0 5 92.56 �0.44Kadum 147.0 2 146.95 �0.05Shintin 4 83.0 3 82.92 �0.08Karne Shomron 52.0 1.5 54.99 2.99Azun 2 35.0 0.5 35.82 0.82Abu Hajla 3 34.0 0.5 34.08 0.08Dir Sharaf 2a 80.0 5 79.86 �0.14Ya’bed 2 25.0 2 27.03 2.03Ariel 285.0 3 286.23 1.23

Lowland Tel Hasi 26.1 0.2 25.94 �0.16

0.8

Amazia 1 26.2 0.2 26.05 �0.15Ayalon 1 24.0 0.1 24.42 0.42Petah Tiqwa oil 23.0 0.1 21.91 �1.09Hadasim oil 21.6 0.2 21.55 �0.05Kakun 1 20.0 0.2 20.97 0.97Ma’anit 2 19.7 0.1 20.37 0.67Barka’I 3 16.9 0.2 17.74 0.84Gal-Ed 3 16.0 0.5 16.96 0.96Yoqneam 4 15.4 1 17.00 1.60Negba 5 25.7 0.3 25.76 0.06Menashe T/1 18.6 0.2 19.01 0.41Menashe T/2 19.2 0.2 19.46 0.26Be’erotaim 1 20.8 0.1 21.44 0.64Kefar Has B 21.6 0.1 21.74 0.14Gezer 1 24.5 0.1 24.62 0.12Mahne Yehuda 23.8 0.1 22.09 �1.71Givat HaShlosha 22.8 0.1 20.88 �1.87Kefar Saba 8 22.3 0.1 21.73 �0.57Ramat HaKovesh B 21.8 0.1 21.77 �0.03Tel Zafit 25.5 0.3 25.51 0.06Kefar Uriyya 4 25.0 0.3 25.01 0.04

270 E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275

inim outlet area. All these values are in agreement with previousstudies of both, the YT basin and adjacent areas (Guttman andKronbuter, 2007; Laronne Ben-Itzhak and Gvirtzman, 2005 andstudies summarized in Table 2). The shoreline of the Taninim blockhowever, was assigned low conductivity (�2.5 m/day, #41 atTable 6), in order to meet the water balance estimations(2.5–3.5 MCM/year).

The vertical conductivity of the aquitard units, at places wheretwo phreatic sub-aquifers were modeled, was found to be about

8 � 10�6 m/day with anisotropy of 1:1000. These low values arelinked to the marly lithology of the Bet-Me’ir and Moza formationsat the Samaria (#132 and #135 in Fig. 9B) and the Soreq Formationat the Hebron Mountains.

2.5. Sensitivity analyses of hydraulic conductivities

In order to evaluate the models sensitivity to changes in thehorizontal hydraulic conductivity, a systematic procedure of local

0

100

200

300

400

500

600

700

Measured heads [m]

Sim

ulat

ed h

eads

[m]

0

10

20

30

0 100 200 300 400 500 600 700

0 10 20 30

Fig. 8. Simulated vs. measured heads for the entire basin. Heads of up to 30 m,located mostly at the confined part, are enlarged.

A

A1

-1.6 >

-1.5 - -1.0

-1.0 - -0.5

-0.5 - 0.0

Tan

Y

Fig. 9. Hydraulic conductivity (m/day) of: (A) the upper sub-aquifer; (B) the mid aquitarfocus at Jerusalem Mountains; C1 – focus to the Taninim Springs. Numbers indicate pol

E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275 271

sensitivity analysis was initiated. Conductivity was altered at oneplace at a time, and the resulting heads and discharges were re-corded. In most cases, the changes in the conductivity lead to slighthead adjustment at nearby regions but not to the entire domain.For example, changes made within the Jerusalem Mountains af-fected only the nearby phreatic wells, and changes made in thecentral lowland affected only the down-gradient confined wells.Thus, the RMSE for each test was comprised of different key-wellsat which head shifts were noted. Acceptable results were limited to100% change of RMSE and 5% change of discharge, in respect to cal-ibrated values.

It was found that changes beyond 15–35% for most areas, in thehydraulic conductivities of the aquifer units affect the distributionof hydraulic head (Tables 4–7). Thus, we are quite confident regard-ing the calibrated hydraulic conductivities of the aquifer units.

On the other hand, considerable changes in the hydraulic con-ductivity and the vertical anisotropy of the allegedly aquitard atthe confined part, were found completely insensitive (Table 8).For example, at the calibrated model head gradient across theintermediate layer was negligible in most of the area, reachingno more than 5 cm in the northern part of the basin. Higher anisot-ropy (1:10,000) resulted differences of up to 20 cm in the north,but did not change the discharges or the overall flow field. Thesecalculated head differences are smaller than the uncertainty ofthe field measurements (Table 3). This testifies the computational

B C

C1

-1.6 >

-1.5 - -1.0

-1.0 - -0.5

-0.5 - 0.0

inim

d; (C) the lower sub-aquifer. Hydraulic conductivities are logarithmic scaled. A1 – aygons as appear at Tables 4–8.

Table 4Sensitivity analyses of hydraulic conductivity at the Jerusalem Mountains.

# Location Calibrated K(m/day)

8 En KaremWellsa

Calibrated RMSE (m) K range [to satisfy average head error below twice calibrated RMSE]

(m/day) ±Percent

17 Syncline axis West 6.739 + 5.728–8.087 �15% to +20%20 Middle 3.801 + 3.497–4.105 �8% to +8%23 East 0.449 + 0.337–0.673 �25% to +50%

18 Hebron anticline flank 0.605 + 0.556–0.695 �9% to +15%

21 Ram’alla anticline axis West 0.086 + 2.5 0.052–0.116 �40% to +35%24 East 0.035 + 0.023–0.049 �35% to +40%

19 Ram’alla anticline flank West 0.441 + 0.287–0.573 �35% to +30%25 East 0.082 + 0.029–0.238 �66% to +190%

a Including En Karem 2, 5, 7, 8, 11, 12, 13 wells and Bar Giora 1.

Table 8Sensitivity analysis of hydraulic conductivity of the Moza and Beit-Me’ir formations at the confined zone.

# Location CalibratedK (m/day)

Gal-ed Menashe T/1 Kakun 1 KefarHasB

PetahTiqwa

Gezer1

TelZafit

TelHasi

BeerSheva1

MashabeySade 1

CalibratedRMSE (m)

Tested K range

1 2 3 4 5 6 7 8 9 10 (m/day) ±Percent

136 North 21.6 + + + + + + + + + + 0.55 0.21–216

�99–1000%127 Middle (west) 23.3 + + + + + + 0.4 0.23–233128 Middle (east) 47.5 + + + + + + + + + + 0.55 0.47–475103 South 19.4 + + + + 0.1 0.94–194

All aquitard polygons simultaneously + + + + + + + + + + 0.55 �97–1000%

Sensitivity analysis ceased when aquitard conductivity reach the assigned conductivity of the above and below sub-aquifers.

Table 5Sensitivity analysis of hydraulic conductivity at the Samaria.

# Location Calibrated K(m/day)

Ya’bed AbuHijla

Azun 2 KarneShomron

DierSharaf

Shibtin 4 Shibtin 1 Kadum Haris Ariel CalibratedRMSE (m)

K range [to satisfyaverage head errorbelow twicecalibrated RMSE]

1 2 3 4 5 6 7 8 9 10 (m/day) ±Percent

33 West 7.3 + + + + + + + + + + 1.7 6.20–12.04 �15% to +65%34 Middle (north) 2.41 + + + + + 2.2 1.69–10.80 �30% to +350%31 Middle (center) 0.86 + + + + + + + + + 1.7 0.69–3.01 �20% to +250%35 l Slopes (NW) 0.52 + + + + + 2.2 0.39–0.75 �25% to +45%30 l Slopes (W) 0.432 + + + + + + + 1.9 0.39–0.49 �10% to +15%32 l Ridge 0.19 + + + + + + + 1.9 0.17–0.22 �10% to +17%30 u Slopes 0.432 + + + + + + 1.7 0.40–0.54 �7% to +26%32 u Ridge 0.19 + + 2.8 0.04–0.65 �80% to +240%

Table 6Sensitivity analysis of hydraulic conductivity at the northern edge of the YT basin.

# Location Calibrated K(m/day)

TaninimT/5

TaninimT/3

Binyamina 1 Shuni 2 Binyamina 5 Tireli 6 Gal-Ed 3 MenasheT/1

CalibratedRMSE(m)

TaninimDQ = 5MCM

InterfaceDQ = 0.5MCM

K range [to satisfyaverage headerror below twicecalibrated RMSE]

1 2 3 4 5 6 7 8 (m/day) ±Percent

39 Taninimblock

1987 + + + + + + + 0.55 + 1570–2543 �21% to +28%

42 Shuni-Tirelibasin

363 + + + + + + + 0.52 + 290–490 �20% to +35%

41 Shore 2.59 + 2.20–2.98 �15% to +15%

Table 7Sensitivity analysis of hydraulic conductivity of the sub-aquifers at the confined zone.

# Location Calibrated K(m/day)

Gal-ed MenasheT/1

Kakun 1 KefarHas B

PetahTiqwa

Gezer 1 TelZafit

TelHasi

BeerSheva 1

MashabeySade 1

CalibratedRMSE (m)

K range [to satisfyaverage headerror below twicecalibrated RMSE]

1 2 3 4 5 6 7 8 9 10 (m/day) ±Percent

36 North 216 + + + + + + + + + + 0.55 162–260 �25% to +20%27 Middle (west) 233 + + + + + + 0.4 151–349 �35% to +150%28 Middle (east) 475 + + + + + + + + + + 0.55 356–665 �25% to +40%3 South 194 + + + + 0.1 146–232 �25% to +20%

272 E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275

E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275 273

model’s limitation in determine the hydrostratigraphy of the JGA,and the need in conceptual assumption and further data to addressthe issue.

3. Discussion

3.1. Structure-dictated flow

Indeed, the 3D numerical model exhibits the influence of thegeological structure on the flow field: At the extensively-folded Ju-dea Mountains, most groundwater is diverted south-westward,parallel to the synclinal axes, leaving minor quantities to followthe general E–W hydraulic gradient. The latter occurs at placeswhere water table is slightly higher than the adjacent anticlinecrests (Fig. 7).

Groundwater flow across the anticline flanks is enabled byassigning zones of low hydraulic conductivity along the anticlinecrests, as against hydraulic barrier in previous models (Baidaet al., 1978; Shachnai, 1980; Meiri and Guttman, 1984; Baida andZukerman, 1992). However, low hydraulic conductivity alone isnot sufficient to maintain the steep gradient across the anticline.Rather, it is maintained by a constant, small ‘overflow’ from theelevated internal sub-basins. This setting occurs along the westernflank of the Ram’alla and Hebron anticlines (Fig. 7A and B) and theeastern flanks of the Ma’on and Zohar anticlines (Fig. 3B). Suchoverflow does not occur at the northern part of Ram’alla anticline(Fig. 7C), hence the JGA at the Binyamin Mountains is unsaturated(dry).

The structural influence is not exclusive to the phreatic parts;the diverted flow paths reach the confined zone through two ‘en-trances’, narrow zones located at the western edge of each syncli-nal axis: the southern Tel-Shoket entrance, down-gradient theHebron Mountains and the northern Ajur entrance, down-gradientthe Jerusalem Mountains (see locations at Fig. 1). In other words, atthe lowland, flow-lines converge from several sub-basins, and thegroundwater flow is intensified. Each flow-line pushes the previ-ous flow-lines westward (Fig. 5).

3.2. Spatial distribution of the hydraulic conductivity

Based on the calibration procedure and the sensitivity analyses(summarized in Tables 4–8), we obtained the best-fitted hydraulicconductivity value for each of the polygons composing the aquifermesh, as is shown in Fig. 9. Two zones are identified. The first com-posed of the confined part of the basin and the foothills. The secondcomposed of the elevated mountains region. These zones matchthe ‘karstified area’ and the ‘slightly karstified area’ defined byFrumkin and Fischhendler (2005). At the ‘karstfied area’, horizontalhydraulic conductivity is relatively high and rather homogeneous(2–5 � 102 m/day, Table 7). At the ‘slightly karstified area’, on theother hand, the hydraulic conductivity zoning is linked with thegeological folds. The general phenomenon observed is that asgroundwater flow quantity increases, the hydraulic conductivityalso increases. We interpret this result by the karstification mech-anism. Thus, where groundwater flow-lines converge and wheregroundwater discharge amount increases, the karstification pro-cess intensifies and permeability increases. Consequently, alongthe syncline axes, where groundwater flow-lines converge, higherconductivities are found.

At the Judea Mountains hydraulic conductivities increase from10�2 m/day along the crest of anticlines, to 1 m/day at the sysn-clines (Table 4). Our results stand in good agreement with previouswork regarding Ram’alla anticline (Yechieli et al., 2007). Similarpattern is detected at the Samaria, where hydraulic conductivityincreases gradually westward, from 10�1 m/day along the crest ofAnabta anticline to 1 m/day at its flank (Table 5). Analogous trend

of reduce hydraulic conductivity toward the anticline crest was re-ported also under the Judea Desert, at the eastern part of themountain aquifer (Laronne Ben-Itzhak and Gvirtzman, 2005).

The lateral shift between steep gradients and low conductivi-ties, characterize the mountainous areas, to the very low gradientand the high conductivities at the lowland, is not coincide with thelateral shift between phreatic and confined conditions. The formeris change abruptly at the point where the JGA base dips below thelowland water table (Fig. 7A and B). At this point, the saturatethickness of the aquifer is boost from several tens of meters to al-most 1000 m.

The gradual trend of decreased conductivities towards the anti-cline crests was explained in previous studies to be related to: (1)groundwater flow in oblique angle to the bedding planes, whichact to reduce conductivity due to anisotropy or (2) closure of jointsand voids due to compressional stresses acting on the reverse faultat the base of anticline (Yechieli et al., 2007). However, we herebysuggest a third alternative hypothesis, which links the conductivitydistribution to the karst characteristics:

3.3. Paleo-Karst control over hydraulic conductivity

The JGA exhibits a unique situation in which well-developedkarst conduits exist under confined conditions and less developedunder phreatic conditions. This allegedly misfit the conventionalunderstanding (Ford and Williams, 2007) and required a detaildiscussion.

Currently, and probably during the last hundred thousand ofyears, the JGA at the lowland is confined, and impede furtherdevelopment of karst conduits. This is indicated also by the satura-tion of the water in respect to calcite and dolomite (Bar, 1983; Kro-itoru et al., 1989). Thus, it is hypothesis that the karst conduits atthe lowland had developed during earlier geological periods, inwhich this part of the basin was unconfined. Such periods mustbe associated with relatively lower base level, i.e. drop in sea level.Mediterranean sea level had dropped dramatically in the past dur-ing two periods: the Messinian Salinity Crises (MSC, at 5.6–5.3 Ma,Lofi et al., 2005) and the glaciations periods of the Pleistocene(<0.5 Ma, Rohling et al., 1998).

At the MSC, the sea level had dropped more than 1000 m andthe Mediterranean had desiccated (Lofi et al., 2005). Great canyonscarved the seaward slopes of the governing continents and drainedthe surface water as well as the groundwater (Gvirtzman, 1970,2002; Lofi et al., 2005; Bakalowicz et al., 2008). Bakalowicz et al.(2008) hypothesized that during the MSC multi-phase karst sys-tems had been developed at the Upper Cretaceous calcareous rocksof Lebanon at different levels, following the dropping sea level. Thegeological settings at their study are similar to that of the JGA, thustheir arguments and conclusions seem to be plausible also for theYT basin.

At the Pleistocene, the sea level had dropped several times to atmost 120 m (Rohling et al., 1998). However, at that time the JGAhad been already covered under thick section of Yafo Formationmarls, and the water could not be drained westward, to the sea(for example, Fig. 3B). Lowering the head at the entire basin byno more than 120 m would enhance the karst systems only atthe upper part of the aquifer (see Fig. 3 for general expression).

During these two periods, successive karst system developed atgreater depths, while upper karst systems were abandoned. Suchphenomenon is reported from other karst terrains around theworld (Ford and Williams, 2007). For example, at the Holloch cave,Switzerland, three successive levels were found within a 600 mvertical section (Bogli, 1980). Successive karst development is alsoshown in the study area, but in a smaller scale, at the Ayalon Quar-ry, having several levels over a 50 m vertical section (Frumkin andGvirtzman, 2006).

274 E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275

Inevitably, the karst systems included vertical conduits (shafts,joints and fractures) which drained the water from the surface tothe evolving systems at the level of the drainage base. These verti-cal conduits cut across different parts of the JGA (Frumkin andGvirtzman, 2006). Currently, the deep paleo-karst systems, whichinclude large, open, horizontal and vertical conduits, enhance boththe horizontal and the vertical transmissivity of the JGA. Thereby,the original lithological stratification plays a minor role. Due tothe high transmissivity, the head gradient from south to northand between the upper and lower parts of the JGA, respectively,is extremely low. Thus, separation of the JGA into sub-aquiferson a stratigraphic basis seems inapt. It is hypothesize that the mazecaves at the upper part of the JGA form at the Pleistocene, near thepast water table, in accordance with literature (Ford and Williams,2007). The existence of several abandoned levels at the Ayaloncave reflects regional changes at the water table elevation, or mostprobably structural uplifting. The later was studied extensively byBar et al. (2006) which conclude that a major uplifting phase tookplace at the Pliocene–Pleistocene boundary. The MSC karst systemfound deeper and currently under confined conditions, thus canonly be penetrated sporadically at deep wells.

4. Summary

This work presents a conceptual and numerical analyses of theaccumulating effect of karst, geological folds and lithology on thegroundwater flow field in the thick carbonate JGA beneath thewestern part of Israel.

Lithological effects on flow field result from the initial deposi-tion environment. In the YT basin, lithological stratification is re-stricted to the mountainous recharge areas, where the JGA isdivided into several sub-aquifers, as the interbeded marls formsaquitards. The most distinct vertical division is shown in the Heb-ron Mountains, where four such sub-aquifers exist (Fig. 2B), but lo-cal divisions to sub-aquifers exist also at the Jerusalem andSamaria Mountains. Lithological effects at the confined part arevery limited, as the JGA becomes relatively homogenous and be-cause it is crossed by well-developed vertical and horizontal karstsystems.

Folding effects on the flow field are limited to the unconfinedzones, where the saturate thickness of the aquifer is relatively thin(<200 m, Fig. 7). At the YT basin they are best visible at the JudeaMountainous, which comprises of three prominent anticlines. Awater divide or alternatively a physical boundary is developedupon the exposed anticlines, and thus groundwater diverted obli-que to the anticlines. A major water amount is diverted from thesteep E–W head gradient to a moderate NE–SW head gradient,and flows along the syncline axes. Similar conclusion was alsonoted at the eastern part of the Israeli Mountain aquifer (LaronneBen-Itzhak and Gvirtzman, 2005). This has also subtle effects ongroundwater flow at the confined zone, as large quantities ofgroundwater reaches the confined zone through narrow ‘en-trances’ at the western edge of each synclinal axis, but only smallwater quantities reaches the confined zones through other widersections.

Karst however, is probably the most prominent feature to con-trol groundwater flow field in the JGA, as happens in many othercarbonatic aquifers (Ford and Williams, 2007), for several reasons:(a) it enhance the bulk conductivity of the rocks by several ordersof magnitude (Freeze and Cherry, 1979); (b) it inter-connects dif-ferent parts of the basin and different horizons within the aquiferand balancing the pressures across the horizontal and vertical sec-tions, respectively; and (c) it enables fast recharge of the aquifer.The first two effects occur however only when well-developedkarst systems cross the aquifer, as the case at the confined part

of the YT basin is. The 3D karst systems there are related to pa-leo-hydrological conditions. On the other hand, wherever karstsystems are less developed, i.e. comprises of unconnected 1D con-duits, karst systems merely allow fast intake of the recharge. This isthe case at the unconfined parts of the YT basin.

Accounting the detailed 3D geological structure in the numeri-cal model significantly improves the resulted head map, the con-ceptual understanding and the derivate conclusion. It allowsdelineation of several distinct water tables and to define the con-nections areas in-between. Moreover, it avoids introduction ofarbitrary artificial conditions (such as barriers). Nevertheless, onthe era when computers modeling importance in hydrogeologicalstudy rise, it seems that a proper effort should be investigate intoconceptual characterization of the studied system. Paleo-hydrol-ogy, in particular, may bear an explanation to current phenomenonin the groundwater flow field.

Acknowledgments

The research was funded by the Israeli Water Authority. Theauthors thank Dr. Gabi Weinberger, Dr. Avihai Hadad andDr. Joseph Guttman for fruitful discussions and help.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.jhydrol.2010.05.038.

References

Arkin, Y., Braun, M., 1965. Type sections of Upper Cretaceous formation in thenorthern Negev. Geological Survey of Israel. Stratigraphic Sections 2a.

Arkin, Y., Hamaoui, M., 1967. The Judea Group (Upper Cretaceous) in central andsouthern Israel. Israel Geological Survey Bulletin, vol. 42. Jerusalem, 17p.

Baida, A., Zukerman, H., 1992. Jerusalem area-prospective and exploitation of thewater resources at the Cenomanian aquifer. Rep. 01/92/10. Tahal Ltd., Tel Aviv(in Hebrew).

Baida, A., Goldshtoff, Y., Kidron, Y., 1978. Numerical model of the Cenomanianaquifer at the southern part of Yarqon-Taninim basin (Beer-Sheva basin). Rep.01/78/08. Tahal Ltd., Tel Aviv (in Hebrew).

Bakalowicz, M., El Hakim, M., El-Hajj, A., 2008. Karst groundwater resources in thecountries of eastern Mediterranean: the example of Lebanon. EnvironmentalGeology 54, 597–604.

Bar, Y., 1983. The hydrogeology and geochemistry of ground water in the area ofBinyamina – Umm el-Fahem. M.Sc. Thesis. The Hebrew University of Jerusalemand Israeli Hydrological Service, Rep. 7/83 (in Hebrew).

Bar, O., Zilberman, E., Gvirtzman, Z., Feinstein, S., 2006. Reconstruction of the upliftstages of the mountain backbone in central Israel, Rep. GSI/28/06. TheGeological Survey of Israel, Jerusalem, 73p (in Hebrew) + English Abstract.

Bein, A., 1974. Reef development at the Judea Group Rocks in the Carmel and in theIsraeli coast. Ph.D Thesis. The Hebrew University, Jerusalem (H, EnglishAbstract).

Ben Gai, Y., Fleischer, L., Goldberg, I., Gendler, M., Gafsu, R., Gvirtzman, H., Dafny, E.,Steinberg, J., 2007. Structural and lithofacial structure for hydrogeologicalmodel of the Yarqon-Taninim aquifer. Rep. 425/255/07. Geophysical Institute ofIsrael, Lod, 59p. + Figures. (in Hebrew).

Berger, D., 1999. Hydrological Model for the Yarqon-Taninim Aquifer. Mekorot Ltd..50p (in Hebrew).

Bogli, A., 1980. Karst Hydrology and Physical Speleology. Springer-Verlag, Berlin.284pp.

Diersch, H.J.G., 1996. Interactive, graphics-based finite-element simulation systemFEFLOW for modeling groundwater flow, contaminant mass and heat transportprocesses. User’s Manual v. 4.5, Institute for Water Resources Planning andSystem Research, Ltd., The Netherlands.

Druckman, Y., Buchbinder, B., Martinotti, G.M., Siman-Tov, R., Aharon, P., 1995. Theburied Afiq Canyon (eastern Mediterranean, Israel): a case study of a Tertiarysubmarine canyon exposed in Late Messinian times. Marine Geology 123, 167–185.

Fleischer, L., Gelbermann, E., Wolff, O., 1993. A geological–geophysical reassessmentof the Judeoa Group (Yarqon-Taninim aquifer). Rep. 244/147/92. Institute forPetroleum research and geophysics, Lod, 92p. + Maps.

Ford, D., Williams, P., 2007. Karst Hydrogeology and Geomorphology. Wiley,England. 562 pp.

Freeze, R.A., Cherry, J.A., 1979. Groundwater. Prentice Hall Inc., New Jersey. 604 pp.Freeze, R.A., Witherspoon, P.A., 1967. Theoretical analysis of regional groundwater

flow: 2. Effect on water-table configuration and subsurface permeabilityvariation. Water Resource Research 3 (2), 623–634.

E. Dafny et al. / Journal of Hydrology 389 (2010) 260–275 275

Frumkin, A., Fischhendler, I., 2005. Morphometry and distribution of isolated cavesas a guide for phreatic and confined paleohydrological conditions.Geomorphology 67, 457–471.

Frumkin, A., Gvirtzman, H., 2006. Cross-formational rising groundwater at anartesian karstic basin: the Ayalon saline anomaly, Israel. Journal of Hydrology318, 316–333.

Garfunkel, Z., 1988. The pre-quaternary geology of Israel. In: Yom-Tov, Y., Tchernov,E. (Eds.), The Zoogeography of Israel. Dr. W. Junk Publishers, Dordrecht,Netherlands, pp. 7–34.

Goldshtoff, Y., Shachnai, E., 1980. Yarqon-Taninim-Beer Sheva basin. Setting andcalibrating numerical model. Rep. 01/80/58. Tahal Ltd., Tel Aviv, 25p. + Figures(in Hebrew).

Guttman, J., 1991. Availability to increase pumping at the Beer-Sheva basin. Rep. 01/91/08. Tahal Ltd., Tel Aviv, 31p (in Hebrew).

Guttman, J., 2002. Exploitation of brackish water for desalination at Or-‘Akiva –Caesarea area. Rep. 756. Mekorot Ltd., Tel Aviv, 9p (in Hebrew).

Guttman, J., Gotleib, M., 1996. Hebron boreholes 1 and 2 – Final report. Rep. 5477-R96.253(E). Tahal Ltd., Tel Aviv, 15p (in Hebrew).

Guttman, J,. Kronbuter, L., 2007. The southern coastal plain of the Carmel-affects ofincreasing exploitation upon local flow regime. Rep. 1291. Mekorot Ltd., Tel-Aviv (in Hebrew).

Guttman, J., Zeitoun, D., 1996. Flow and salinity model for the northern Yarqon-Taninim basin. Rep. 6101-D96.382. Tahal Ltd., Tel-Aviv.

Guttman, J., Zukerman, H., 1995. Yarqon-Taninim – Beer Sheva groundwater basin:setting and calibrating flow model. Rep. 01/95/72. Tahal Ltd., Tel Aviv, 37p (inHebrew).

Gvirtzman, G., 1970. The Saqiye group (Late Eocene to Early Pleistocene) in thecoastal plain and Hashphela regions, Isarel. Israel Geological Survey Bulletin 51.

Gvirtzman, H., 2002. Israel Water Resources: Chapters in Hydrology andEnvironmental Sciences. Yad Ben Zvi press, Jerusalem. 287pp (in Hebrew).

Hydrological Service of Israel, 2007. Annual hydrological report (Hydrological year2005/2006), Jerusalem (in Hebrew).

Kolton, J., 1973. The Geology of the Hebron area. Explanation for the geological map.Rep. 01/73/09. Tahal Ltd., Tel Aviv (in Hebrew).

Kroitoru, L., Carmi, I., Mazor, E., 1989. Groundwater 14C activity as affected byinitial water–rock interactions in the Judean Mountains, Israel. ChemicalGeology 79, 259–274.

Laronne Ben-Itzhak, L., Gvirtzman, H., 2005. Groundwater flow along and acrossstructural folding: an example from the Judean Desert, Israel. Journal ofHydrology 312, 51–69.

Lewy, Z., 1991. Periodicity of Cretaceous epeirogenic pulses and the disappearanceof the carbonate platform facies in Late Cretaceous times (Israel). Israel Journalof Earth Sciences 40, 51–58.

Lofi, J., Gorini, C., Berne, S., Clauzon, G., Dos Reise, T., Ryan, W., Stecklerf, M., 2005.Erosional processes and paleo-environmental changes in the Western Gulf ofLions (SW France) during the Messinian Salinity Crisis. Marine Geology 217, 1–30.

Mandel, S., 1961. Properties and genesis of the Turonian–Cenomenian aquifer inwestern Israel as an example of large limestone aquifer. Ph.D. Thesis. Technion,Haifa (in Hebrew).

Matmon, D., 1995. Simulations of groundwater flow between the MediterraneanSea and the Jordan Rift Valley. M.Sc. Thesis. The Hebrew University ofJerusalem, 157p.

Meiri, D. and Guttman, J., 1984. Reconstruction by flow model of the flow pattern ofthe Jerusalem effluents that leak into the Judea Group aquifer in the SoreqWadi. Rep. 01/84/52. Tahal Ltd., Tel Aviv (in Hebrew).

Mercado, A., 1980. The groundwater salinity in the Yarqon-Taninim-Beer Shevabasin – conceptual model of the salinity regime and sources. Rep. 01/80/60.Tahal Ltd., Tel Aviv, 65p (in Hebrew).

Paster, A., Dagan, G., Guttman, J., 2005. The salt-water body in the northern part ofYarqon-Taninim aquifer: field data analysis, conceptual model and prediction.Journal of Hydrology 323 (1), 154–167.

Rimmer, A., 2003. Groundwater potential under the southern coastal plain andsouthern lowland-hydrological model. Internal Report, 35p (in Hebrew).

Rohling, E.J., Fenton, M., Jorissen, F.J., Bertrand, P., Ganssen, G., Caulet, J.P., 1998.Magnitudes of sea-level lowstands of the past 500,000 years. Nature 394, 162–165.

Sass, E., Katz, A., 1982. The origin of platform dolomite: new evidence. AmericanJournal of Science 282, 1184–1213.

Shachnai, E., 1980. Yarqon-Taninim-Beer Sheva basin. Updating the hydrogeologicalmodel. Rep. 01/80/12. Tahal Ltd., Tel Aviv, 59p. + Figures (in Hebrew).

Weinberger, G., Rosenthal, E., Ben-Zvi, A., Zeitoun, D.G., 1994. The Yarqon-Taninimgroundwater basin, Israel hydrogeology: case study and critical review. Journalof Hydrology 161, 227–255.

Weiss, M., Gvirtzman, H., 2007. Estimating ground water recharge using flowmodels of perched karstic aquifers. Ground Water 45, 761–773.

Yechieli, Y., Kafri, U., Wollman, S., Lyakhovsky, V., Weinberger, R., 2007. On therelation between steep monoclinal flexure zones and steep hydraulic gradients.Ground Water 45 (5), 616–626.